1. Field of the Invention
The present invention relates generally to the field of optical imaging and more particularly to reduction of the peak power and speckle contrast for bright field and dark field inspection applications.
2. Description of the Related Art
Many optical systems are designed to produce images such as an inspection system for a partially fabricated integrated circuit or a photomask. Techniques and apparatus for inspecting circuits or photomasks are well known in the art and are embodied in various commercial products such as many of those available from KLA-Tencor Corporation of San Jose, Calif. Most optical imaging systems use a continuous illumination source. However, many times pulsed illumination sources are preferred or are the only sources available. This is especially true in the DUV spectral region below 250 nm where very few high brightness illumination sources exist that are not pulsed. Common examples are excimer lasers used in the photolithography process for manufacturing semiconductor devices.
If a pulsed illumination source is used, the optical imaging system must contend with the nature of pulsed illumination. This is especially true when inspecting integrated circuits or photomasks. Pulsed illumination typically suffers from two major problems. First, the peak power of the illumination transmitted from the illumination source can be very high and potentially damage elements in the optical system or the object being inspected. Second, the light energy can suffer from xe2x80x9cspecklexe2x80x9d or random intensity distribution of light due to interference effects. This is especially true for laser light sources.
Further, instances occur wherein a higher repetition rate source is not available. A system or device for turning a high repetition rate laser, such as a mode locked laser, into a virtually continuous source would be very useful in these situations.
One prior apparatus for reducing the peak power of a pulsed laser is U.S. Pat. No. 5,309,456 by Horton. The Horton design uses one mirror and one beamsplitter to split a single laser pulse into multiple pulses. The multiple pulses are then delayed with respect to each other using reflective optical delay schemes. Several drawbacks exist for this approach. First, the pulse-to-pulse uniformity is highly dependent on the quality of the mirror and beamsplitter used to form the multiple pulses as well as losses in the optical delay schemes. To maintain uniform pulses this system requires 100% reflective mirrors, 50% reflective and 50% transmissive beam splitters with no absorption, and perfect AR coatings with 100% transmission. Any deviations from this will cause an energy variation between the pulses. For example, consider the effects of imperfect optics on a system that generates 16 pulses. If the beamsplitter transmission is 49% and the reflectivity is 51%, the energy variation between the pulses will be 16%. In addition, if the mirror has a reflectivity of 99% it will cause an additional energy variation between the pulses of 3%. Another limitation of the Horton design is that it is not well suited for the DUV-VUV spectral range. Reflective coatings are much less efficient in this range and can cause large losses. These losses will contribute to pulse-to-pulse nonuniformity and a reduction in the efficiency of the peak power reduction scheme. For example, a reflective optical delay scheme with a 1 m long cavity, using mirrors with 99% reflectivity, and an optical delay of 10 meters will have a loss of 10%. Similarily, a reflective optical delay scheme with an optical delay of 20 m, 40 m, and 80 m will have losses of 18%, 33%, and 55% respectively. If these delay paths are used in a system to generate 16 pulses, assuming perfect 50% beam splitters and a perfectly reflecting mirror, the lowest energy pulse will be only 22% of the highest energy pulse. An additional limitation of the Horton design is that it uses a single mirror and beamsplitter to generate the multiple pulses. This optical setup is not flexible and inhibits compensation of different losses for each delay path. In addition, this scheme offers no solution for dealing with the effects of speckle.
With respect to speckle problems, two primary techniques have been used in the past to reduce the contrast of speckle in a single laser pulse. The first technique employs two rotating diffusers to create multiple speckle patterns during a single pulse. This technique relies on the relative motion of the two rotating diffusers to produce uncorrrelated speckle patterns. This technique has several major disadvantages. First, the diffusers must rotate at a high at a high rate of speed to produce smoothing within a pulse. For a typical pulse of 20 ns, only a limited number of uncorrelated speckle patterns can be produced. Also, losses from diffusers are typically very high. A typical transmission for such a diffuser is 40%. The diffusers in combination then have a transmission of only 16%. In addition, rotating diffusers can be a source of vibration that can effect the image quality of the system. The second technique uses two diffraction gratings and an electro-optic modulator to produce speckle smoothing within a single pulse. This scheme was developed to minimize speckle problems for laser fusion systems. This technique has several limitations including large size and very high cost. In addition, electro optic modulators operating at high bandwidths in the DUV and VUV ranges are not available.
It is therefore an object of the current invention to provide a system or arrangement that can reduce the peak power of a laser pulse emanating from an energy source.
It is another object of the current invention to provide an illumination solution that does not suffer excessive losses due to mirrors, beamsplitters, and optical delay lines but that can produce substantially uniform pulses.
It is a further object of the current invention to provide an illumination solution that can be readily reconfigured while producing optical delays with minimum optical losses, particularly in the DUV-VUV spectral region.
It is still a further object of the current invention to provide an illumination solution, having reduced speckle contrast for a single energy pulse.
It is yet a further object of the current invention to provide for speckle contrast reduction in an illumination system preferably employing a pulsed illumination source wherein said speckle contrast reduction may be employed in combination with other speckle reduction schemes to further reduce the speckle contrast of a single pulse.
It is yet another object of the current invention to effectively increase the repetition rate of a pulsed source and further to achieve quasi-continuous operation from a high repetition rate source.
The present invention is a system and method for reducing the peak power of a laser pulse. The system and method disclosed herein utilize multiple paths using a unique design to divide a pulse received from a light generating device, such as a laser, into multiple lower energy pulses, and delay those pulses such that they may strike the target surface at different times. The design provided herein comprises a plurality of beamsplitters combined with a plurality of delay elements to delay a pulse or pulses transmitted from the light emitting device in an advantageous manner. The design provides the ability to readily divide the pulse into two, four, eight, or conceivably any number of components with components delayed relative to one another. The energy in each pulse can be adjusted using a variety of optical attenuation schemes to produce pulses with substantially uniform energies. Energy received from the light generating device may be split into components using beamsplitters and directed through different paths toward the target, such as a semiconductor wafer surface. Certain optical delay arrangements using prisms, Brewster""s angle surfaces, and reflecting devices employing mirrors or Total Internal Reflection (TIR) surfaces provide delay compensation for the optical paths. These delay schemes can be in classical arrangements such as a White Cell or Herriott Cell or other novel delay schemes described herein.
The system and method further include a design for reducing speckle contrast, wherein a similar arrangement to that presented for the peak power reduction is employed, using beamsplitters, mirrors, and optical delay arrangements, to reduce the contrast of speckle in a single laser pulse. The reduction in contrast is performed based on the fact that laser beams entering a diffuser at a different angle or position produces a changed speckle pattern leaving the diffuser. Multiple speckle patterns may therefore be generated by multiple beams operating at multiple angles or positions through a diffuser, and the speckle patterns may be integrated together to reduce contrast. However, the speckle pattern must arrive at the detector at slightly different times. Thus the design presented herein to reduce peak power may be used with altered angles between the optical paths such that the split or divided light energy components strike the diffuser at suitably different angles.
An alternate embodiment for reducing speckle contrast is disclosed wherein a single pulse is passed in an angular orientation through a grating to create a delayed portion of the pulse relative to the leading edge of the pulse. One side of the pulse is delayed with respect to the other side of the pulse. If this time delay is suitably longer than the coherence length of the laser pulse, multiple zones are created across the pulse that will not interfere. Each of these zones can then pass through a diffuser at different angles and the speckle contrast can be reduced. A second grating can also be used in combination with the first grating to remove the spectral dispersion while maintaining the optical delay from one side of the pulse to the other.
In addition, these techniques to reduce the speckle contrast can be used in combination with other speckle reduction techniques to further reduce the speckle. Two examples of such techniques are a light pipe and a lens array. A light pipe or lens array spatially divides an input beam into multiple beamlets. Each of these beamlets then overlaps at the output of the light pipe or lens array. If the spatial or temporal coherence of the input pulse is sufficient so that one beamlet does not interfere with another, speckle contrast can be reduced.
Further, the method described herein for creating multiple pulses from a single pulse effectively increases the repetition rate of a repetitively pulsed source. The method described herein for reducing the speckle contrast from a single pulse using a grating to delay one side of a pulse with respect to the other side effectively increases the pulse length in time. Using both of these techniques in combination may produce a continuous or nearly continuous source from a high repetition rate source.